Overexpression of Dlx2 Enhances Osteogenic Differentiation of Bmscs and MC3T3-E1 Cells Via Direct Upregulation of Osteocalcin and Alp

International Journal of Oral Science www.nature.com/ijos

ARTICLE OPEN Overexpression of Dlx2 enhances osteogenic differentiation of BMSCs and MC3T3-E1 cells via direct upregulation of Osteocalcin and Alp

Jianfei Zhang1, Wenbin Zhang1, Jiewen Dai1, Xudong Wang1 and Steve Guofang Shen1

Genetic studies have revealed a critical role of Distal-homeobox (Dlx) genes in bone formation, and our previous study showed that Dlx2 overexpressing in neural crest cells leads to profound abnormalities of the craniofacial tissues. The aim of this study was to investigate the role and the underlying molecular mechanisms of Dlx2 in osteogenic differentiation of mouse bone marrow stromal cells (BMSCs) and pre-osteoblast MC3T3-E1 cells. Initially, we observed upregulation of Dlx2 during the early osteogenesis in BMSCs and MC3T3-E1 cells. Moreover, Dlx2 overexpression enhanced alkaline phosphatase (ALP) activity and extracellular matrix mineralization in BMSCs and MC3T3-E1 cell line. In addition, micro-CT of implanted tissues in nude mice conﬁrmed that Dlx2 overexpression in BMSCs promoted bone formation in vivo. Unexpectedly, Dlx2 overexpression had little impact on the expression level of the pivotal osteogenic transcription factors Runx2, Dlx5, Msx2, and Osterix, but led to upregulation of Alp and Osteocalcin (OCN), both of which play critical roles in promoting osteoblast maturation. Importantly, luciferase analysis showed that Dlx2 overexpression stimulated both OCN and Alp promoter activity. Through chromatin-immunoprecipitation assay and site-directed mutagenesis analysis, we provide molecular evidence that Dlx2 transactivates OCN and Alp expression by directly binding to the Dlx2-response cis-acting elements in the promoter of the two genes. Based on these ﬁndings, we demonstrate that Dlx2 overexpression enhances osteogenic differentiation in vitro and accelerates bone formation in vivo via direct upregulation of the OCN and Alp gene, suggesting that Dlx2 plays a crucial role in osteogenic differentiation and bone formation.

International Journal of Oral Science (2019) 11:12 ; https://doi.org/10.1038/s41368-019-0046-1

INTRODUCTION development of the primordium, as it leads to the development of The distal-less homeobox (Dlx) gene family consists of six the maxillofacial skeletal pattern.6 Given that Dlx5 controls members (Dlx1, Dlx2, Dlx3, Dlx5, Dlx6, and Dlx7); these members osteogenic differentiation,7 it is reasonable to speculate that are expressed in a complex pattern in the first and second Dlx2 might be involved in this process. So far, only a few studies branchial arch region.1 Dlx1/2 regulates the development of the have reported that Dlx2 overexpression increases the osteogenic upper jaw, while Dlx5/6 regulates lower jaw development.2 differentiation potential of pre-osteoblast cells.8 However, how Together with other homeobox proteins, the Dlx family regulates Dlx2 regulates osteogenic differentiation and the underlying osteoblast differentiation. As one of the key transcription factors cellular and molecular mechanisms remain unknown. regulating osteogenic differentiation, Dlx5 stimulates two other In a previous study, we found that elevated Dlx2 expression led key transcription factors, Runx2 and Osterix (Osx), which to midfacial development defects, nasal deformities, premaxillary sequentially induce expression of bone markers such as Osteo- bony deficiency, and spine deformities.9 Thus, it is crucial to calcin (OCN) and Alkaline Phosphatase (ALP). The expression of Dlx5 examine how Dlx2 overexpression leads to abnormal bone is induced by bone morphogenetic protein-2 (BMP-2).3 Msx2, formation both in vitro and in vivo. To investigate the role of another homeobox gene and a key regulator of osteogenic Dlx2 during osteogenic differentiation both in vitro and in vivo, we differentiation, represses the expression of Alp by directly binding used mouse bone marrow stromal cells (BMSCs) in our study, as the to its promoter, while Dlx5 activates its expression by interfering ability of BMSCs to differentiate toward adipogenic, chondrogenic, with the ability of Msx2.4 Thus, Dlx5 coordinates with Msx2 to and osteogenic cell lineages has been characterized extensively regulate osteogenic differentiation due to their reciprocal ability in vivo and in vitro by various researchers.10 Osteogenic to compete with each other. differentiation of BMSCs can be assayed in vitro by ALP and Sharing strong sequence similarity with Dlx5, Dlx2 has been Alizarin red staining and in vivo by transplantation assays.11,12 shown to play a crucial role in craniofacial skeletal development.5 Therefore, mouse BMSCs are suitable for investigating the effect of Dlx2 is upregulated in the central area of the first branchial arch Dlx2 overexpression on osteogenesis both in vitro and in vivo. during days 9.5 and 10.5 of embryonic development in mice. This Murine osteoblastic cell line MC3T3-E1 cells were also chosen to upregulation of Dlx2 is important for the differentiation and verify the effect of Dlx2 overexpression on osteogenesis in vitro.

1Department of Oral & Cranio-maxillofacial Surgery, Shanghai Ninth People’s Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai Key Laboratory of Stomatology, No. 639 Zhizaoju Road, Shanghai, China Correspondence: Xudong Wang ([email protected]) or Steve Guofang Shen ([email protected]) Received: 7 November 2018 Revised: 2 January 2019 Accepted: 3 January 2019 Dlx2 transactivates OCN and ALP to induce osteogenesis Zhang et al. 2 Initially, we observed the upregulation of Dlx2 in both mouse control, parallel BMSCs or MC3T3-E1 cells were transduced with BMSCs and MC3T3-E1 cells during osteogenic differentiation. Lenti-CTRL lentivirus as mock control. Dlx2 expression was then Moreover, forced overexpression of Dlx2 led to enhanced evaluated by RT-qPCR and western blot analysis. Lenti-DLX2 OE- osteogenic differentiation potential of both BMSCs and MC3T3- transduced BMSCs (over) displayed much more Dlx2 transcripts E1 cells in vitro, and accelerated bone formation in vivo. These (Fig. 1c) and correspondingly increased protein production findings prompted us to explore the underlying mechanisms. To (Fig. 1d) than wildtype or Lenti-CTRL-transduced BMSCs (control) our surprise, we found that Dlx2 overexpression had no significant did. Similarly, Lenti-DLX2 OE transduction in MC3T3-E1 cells also effect on the expression levels of Dlx5, Msx2, Runx2, and Osx, but led to increased mRNA and protein levels of Dlx2 (Fig. 1c, d). These led to upregulation of Alp and OCN in BMSCs and MC3T3-E1 cells. observations indicate that Dlx2 was successfully overexpressed in Considering the fact that Alp promotes the early stage of both cell lines. osteogenic differentiation and OCN accelerates the late stage, we next analyzed the promoter of OCN and Alp through luciferase- Dlx2 overexpression enhances the osteogenic differentiation reporter assay and chromatin-immunoprecipitation (ChIP) analysis, potential of BMSCs and MC3T3-E1 cells and found that Dlx2 transcriptionally regulated OCN and Alp To investigate whether the overexpression of Dlx2 effects on expression by directly binding to their promoters. Taken together, osteogenic differentiation in vitro, we carried out ALP staining and our data demonstrates for the first time that Dlx2 overexpression Alizarin staining assays. Lentivirus-transduced BMSCs or MC3T3-E1 enhances the early stage of osteogenic differentiation via direct cells were cultured in OIM for 7, 14, or 21 days to assess their upregulation of Alp, and promotes the late stage of osteogenic osteogenic differentiation potential.14 Interestingly, ALP staining differentiation via direct upregulation of OCN. was significantly enhanced at day 7 and 14 in the Dlx2- overexpressing BMSCs compared with that in control BMSCs (Fig. 2a, c). Similarly, higher ALP activity levels were observed in RESULTS Dlx2-overexpressing BMSCs at day 14 after osteogenic induction Endogenous Dlx2 expression in BMSCs and MC3T3-E1 cells during (Fig. 2b). Furthermore, Alizarin red staining revealed that osteogenesis mineralization was markedly enhanced in Dlx2-overexpressing First, we examined the levels of Dlx2 expression upon osteogenic BMSCs during the entire culture period, especially at day 21 induction in mouse BMSCs and MC3T3-E1 cells. Quantitative (Fig. 2a, c). Consistently, we also observed enhanced ALP and reverse transcription polymerase chain reaction (RT-qPCR) results Alizarin red staining in Dlx2-overexpressing MC3T3-E1 cells (Fig.

1234567890();,: showed that when BMSCs were exposed to osteogenic-inducing S1). Given that ALP activity is involved in the early stage of medium (OIM), Dlx2 expression was upregulated within 0.5 and 3 osteogenic differentiation while mineralization is involved in the h after induction (Fig. 1a). However, after 7- or 14-day culture in late stage, we therefore proposed that Dlx2 overexpression OIM, these cells express similar mRNA level of Dlx2 with the cells accelerates the early stage of osteogenesis by increasing ALP cultured in normal culture medium (data not shown). In addition, activity and the late stage of osteogenesis by enhancing western blot analysis with an anti-Dlx2 antibody detected only a mineralization. very weak signal of Dlx2 protein in BMSCs cultured both in normal medium and OIM for 3 h. This could be explained by the low Forced overexpression of Dlx2 in BMSCs accelerated bone protein level of endogenous Dlx2 in BMSCs. Supporting this formation in vivo notion is the finding that the endogenous protein level of other To investigate whether Dlx2 overexpression could affect osteo- Dlx proteins, such as Dlx5, is also quite low in BMSCs.13 genesis in vivo, we subcutaneously implanted the BMSCs/β- The expression pattern of Dlx2 in MC3T3-E1 cells was similar tricalcium phosphate (β-TCP) constructs into nude mice. The with that in BMSCs. The mRNA level of Dlx2 in MC3T3-E1 cells was whole implanted tissues were then analyzed with microscopic upregulated within and the first 6 h after osteogenic induction computed tomography (micro-CT) examination 6- or 8-weeks (Fig. 1b), but showed no differences with that in the cells cultured after implantation (Fig. 3a). Micro-CT revealed new bone in normal culture medium after culture for 7 or 14 days (data not formation in both control and Dlx2 overexpression groups shown). Consistently, previous study in stem cells from apical (Fig. 3b, c). However, quantitative morphometric analysis showed papilla (SCAP) also showed a similar expression pattern that Dlx2 that bone volume/tissue volume (BV/TV) in the Dlx2-overexpres- was upregulated within the first few hours upon osteogenic sing group (17.81% ± 1.23% and 23.64% ± 1.71% at week 6 and induction, and its expression then showed no significant week 8, respectively) was significantly higher than that in the difference with that in cells grown in normal cell culture medium. control group (12.91% ± 1.16% and 17.04% ± 1.62% at week 6 and This reduction of Dlx2 expression in the induced group could be week 8, respectively) (P < 0.05) (Fig. 3d). Similarly, the Dlx2- explained by the regulation of mRNA stability by microRNAs overexpressing group showed higher bone mineral density (BMD) (miRNAs). Latest study reveals that Dlx2 is a target gene of the of the newly formed bone ((1.477 ± 0.097) g· cm−3 and (1.550 ± miRNA, miR-185-5p, and its expression is negatively regulated by 0.121) g· cm−3 at week 6 and week 8, respectively) than the Dlx2.14 The initial upregulation of Dlx2 may stimulate the control group ((1.055 ± 0.072) g· cm−3 and (1.107 ± 0.098) g· cm−3 expression of downstream target genes, which induces miRNA at week 6 and week 8, respectively) (Fig. 3e). Besides, trabecular expression and sequentially exert a feedback on Dlx2 expression. number (Tb.N) in the Dlx2-overexpressing group ((1.362 ± 0.110) In addition, western blot analysis detects only weak expression of g· cm−3 and (1.713 ± 0.129) g· cm−3 at week 6 and 8, respectively) Dlx2 in MC3T3-E1 cells treated either with OIM or normal culture was significantly higher than that in the control group ((0.979 ± medium, consistent with the endogenous protein level of Dlx2 in 0.086) g· cm−3 and (1.232 ± 0.097) g· cm−3 at week 6 and 8, BMSCs. Together, these results demonstrate that Dlx2 was respectively) (Fig. 3f). Additionally, the Dlx2-overexpressing group upregulated during early osteogenesis in mouse BMSCs and exhibited decreased trabecular space (Tb.Sp) ((0.634 ± 0.059) mm MC3T3-E1 cells. and (0.482 ± 0.033) mm at week 6 and 8, respectively) when compared to the control group ((0.885 ± 0.079) mm and (0.672 ± Forced overexpression of Dlx2 in BMSCs and MC3T3-E1 cells 0.044) mm at week 6 and 8, respectively) (Fig. 3g). Taken together, Next, to investigate the effect of Dlx2 on osteogenesis, we these results demonstrate that overexpression of Dlx2 in BMSCs established Dlx2-overexpressing BMSCs and MC3T3-E1 cells. substantially triggers osteogenic differentiation and improves Cultured BMSCs or MC3T3-E1 cells were transduced with Lenti- bone formation in vivo, defining a pivotal role of Dlx2 in DLX2 OE lentivirus to allow stable Dlx2 overexpression. As a osteogenic differentiation.

International Journal of Oral Science (2019) 11:12 Dlx2 transactivates OCN and ALP to induce osteogenesis Zhang et al. 3 abBMSCs MC3T3-E1 cells ** Normal Normal 4 * Induced 8 * Induced * 3 * 6 * expression expression 2 2 Dl Dl * 2 4

1 2 Relative fold change in fold Relative Relative fold change in fold Relative 0 0 0 min 30 min 1 h 3 h 6 h 24 h 3 d 0 min 30 min 1 h 3 h 6 h 24 h 3 d

c Blank Control Over d ** BMSCs MC3T3-E1 250 ** Blank Control Over Blank Control Over 200 ** 150 ** Dlx2 expression 2 100

Dl β-actin 50 2.0 1.5 1.0 0.5 0.0 Relative fold change in fold Relative MC3T3-E1 cells BMSCs Fig. 1 Analysis of Dlx2 expression by RT-qPCR in BMSCs and MC3T3-E1 cells upon osteogenic induction. Endogenous expression of Dlx2 in BMSCs (a) and MC3T3-E1 cells (b) at different time points upon osteogenic induction. Relative transcript levels of Dlx2 at each time point were quantified by RT-qPCR and normalized with a house-keeping gene Gapdh. Student’s t tests were used to determine statistical significance; n = 3. Error bars represent SDs. *P < 0.05; **P < 0.01. c Dlx2 expression in BMSCs and MC3T3-E1 cells was evaluated with RT-qPCR. Gene expression was normalized with Gapdh, and statistical significance was determined as described in a. d Protein levels of Dlx2 in BMSCs and MC3T3-E1 cells were measured by Western blot analysis. Blank, BMSCs/MC3T3-E1 cells; control, Lenti-CTRL transduced BMSCs/MC3T3-E1 cells; over, Lenti- DLX2 OE transduced BMSCs/MC3T3-E1 cells. β-Actin was used as an internal control

Effect of forced overexpression of Dlx2 on the expression of of Dlx2 in BMSCs leads to upregulation of Alp and OCN, instead of osteogenic genes in BMSCs Runx2, Dlx5, Msx2, and Osx. To gain an insight into the mechanism of Dlx2-induced osteogenesis, we examined the expression of a panel of Dlx2 overexpression leads to increased expression of OCN and Alp osteogenesis-associated genes by RT-qPCR in Dlx2-overexpressing in MC3T3-E1 cells BMSCs and control BMSCs cultured in OIM. As shown in Fig. 4a, RT- We next tested whether Dlx2 overexpression could also lead to qPCR analysis revealed that transduction of Lenti-DLX2 OE into upregulation of OCN and Alp in other osteoblast precursor cell BMSCs resulted in a 526.2-fold and 481.3-fold increase in Dlx2 lines. Compared with mock-transduced cells, we observed expression at days 14 and 21 of osteogenic differentiation, enhanced mRNA and protein levels of OCN in Dlx2-overexpres- respectively. Unexpectedly, there was no difference in Runx2, Dlx5, sing MC3T3-E1 cells after osteogenic induction (Fig. 4h, i). Dlx2- Msx2, and Osx expression between Dlx2-overexpressed BMSCs and overexpressing MC3T3-E1 cells also exhibited higher Alp mock-transduced BMSCs at 14 and 21 days after osteogenic transcription (Fig. 4h) and enhanced ALP activity (Fig. S1A). induction (Fig. 4b–e) nor at 1, 2, and 7 days after osteogenic Moreover, as in BMSCs, we observed no signiﬁcant difference in induction (data not shown), indicating that these genes might not Msx2, Dlx5, Runx2,andOsx expression after Lenti-DLX2 OE be involved in the Dlx2-induced promotion of osteogenesis. transduction in MC3T3-E1 cells after osteogenic induction, as Importantly, forced overexpression of Dlx2 led to upregulation of shown in the Figure S2. OCN at days 14 and 21 after osteogenic induction and Taken together, these data demonstrate that forced over- upregulation of Alp at day 14 (Fig. 4f, g), consistent with the expression of Dlx2 induces upregulation of OCN and Alp upon ALP staining and Alizarin red staining results (Fig. 2a–c). In osteogenic induction both in BMSCs and MC3T3-E1 cell lines. summary, we found, for the ﬁrst time, that forced overexpression Given that the expression levels of Runx2, Dlx5, Msx2, and Osx

International Journal of Oral Science (2019) 11:12 Dlx2 transactivates OCN and ALP to induce osteogenesis Zhang et al. 4

a ALP staining Alizarin red staining Control OverControl Over b 8 ** Day 7 Day 14 6

2 Day 14 Day 21 Relative fold change of fold Relative 0 alkaline phosphatase activity Control Over

c ALP staining Alizarin red staining

×40 ×100 ×200 ×40 ×100 ×200

Control Control

Day 7 Day 14

Over Over

Control Control

Day 14 Day 21

Over Over

Fig. 2 Forced overexpression of Dlx2 enhanced osteogenesis of BMSCs in vitro. a ALP staining was performed on days 7 or 14 after osteogenic induction. Alizarin red staining was carried out after cells were cultured in OIM for 14 or 21 day. b Semi-quantitative analysis of ALP activity in Dlx2-overexpressing BMSCs (over) and control BMSCs (control) after 14-day culture in OIM. c Magnified views of ALP staining and Alizarin red staining in a. Scale bar = 50 μm in all the panels. Left panel, 40-fold magnified image; middle panel, 100-fold magnified image; right panel, 200-fold magnified image

remained unchanged upon Dlx2 overexpression, we speculate Dlx2-FLAG into MC3T3-E1 cells, suggesting that the OCN promoter that OCN and Alp are the direct target genes of Dlx2. contains at least one Dlx2-response element (RE). To further identify the Dlx2-RE(s) in the OCN promoter, we Characterization of the mouse OCN promoter and identification of performed ChIP analysis in MC3T3-E1 cells transfected with pCMV- its Dlx2-response element Dlx2-FLAG. Semi-quantitative PCR analysis showed one strong Previous genetic studies have shown that through direct binding signal in the E6 region (−1311bpto−1 175 bp) matching the to the promoter of OCN, homeodomain (HD) proteins Msx2, Dlx3, predicted ~175 bp size of the E6 PCR product (Fig. 5b). Considering and Dlx5 regulate the expression of OCN in osteogenic cells. Dlx3 the low sensitivity of semi-quantitative PCR, we next carried out RT- binds the OCN promoter to stimulate its expression, while the qPCR. Consistent with the above findings, qPCR also revealed that binding of Msx2 and the recruitment of Dlx5 represses OCN Dlx2 was highly enriched at region E6 (−1311to−1 175 bp) and at expression. Moreover, Dlx5 and Msx2 regulate Alp expression by region E9 (−1073bpto– 932 bp) (Fig. 5c). The qPCR results of directly binding to its promoter.15,16 Considering that Dlx2, an HD region E10 to region E16 are not shown since the signals at these protein, shares strong sequence similarity with Dlx5 and Dlx3, we regions were as weak as the negative control in semi-quantitative speculated that Dlx2 might also be involved in the regulation of PCR results. In addition, the occupancy of Dlx2 in the OCN promoter OCN and Alp transcription in osteogenic cells by binding to their was correlated with increased transcription represented by elevated promoters. occupancy of RNA polymerase II (Pol II) (Fig. 5c), while the negative We first determined whether OCN is directly regulated by Dlx2, control (cells introduced with pCMV-FLAG) showed very weak signal we analyzed its promoter in MC3T3-E1 cells. We inserted the (data not shown). These results indicate that there is a Dlx2-RE in whole promoter region of OCN into the pGL3-basic plasmid, the OCN promoter in region E6 (in primer set 6) and E9 (in primer generating pGL3-OCN, and transferred this vector into MC3T3-E1 set 9). Consistent with this, bioinformatics analysis (JASPAR cells along with pCMV-Dlx2-FLAG to allow the overexpression of database) indicated that the promoter region of OCN contains Dlx2. As shown in Fig. 5a, we observed an ~3.2-fold increase in the potential Dlx2-REs (−1447bpto−1 444 bp and −1113bp transcriptional activity of pGL3-OCN after introduction of pCMV- to −1 110 bp), both of which have the ATTA sequence.

International Journal of Oral Science (2019) 11:12 Dlx2 transactivates OCN and ALP to induce osteogenesis Zhang et al. 5

ab6 weeks BMSCs control BMSCs over C57/BL6 mice mBMSCs

Lenti-CTRL

Lenti-DLX2

1 mm 12.91% 1 mm 17.81%

c 8 weeks TCP/HA BMSCs control BMSCs over

Ostengensis introduction for 7 days 23.64% 1 mm 17.04% 1 mm

d Control Over 0.3 ** ** 0.2

Nude mice subcutaneous implantation BV/TV 0.1

0.0 6 weeks 8 weeks

e f g Control Over Control Over1.5 Control Over 2.0 * 2.0 * * * * 1.5 1.5 1.0 ) ) -1 * -3

1.0 1.0 Tb.Sp /mm Tb.Sp 0.5 Tb.N/(1·mm BMD/(g·cm 0.5 0.5

0.0 0.0 0.0 6 weeks 8 weeks 6 weeks 8 weeks 6 weeks 8 weeks Fig. 3 Dlx2-overexpressing BMSCs accelerated bone formation in vivo. a Schematic diagram of implantation experiments. The whole BMSCs/β- TCP constructs were obtained for micro-CT analysis 6 or 8 weeks after implantation. b, c Representative 3D reconstructed micro-CT results of the BMSCs/β-TCP constructs at weeks 6 and 8 after subcutaneous implantation. Implants were then harvested after 6 (b)or8(c) weeks, and were scanned by micro-CT. Scale bar = 1 μm. Average BV/TV is indicated below. Over, BMSCs transduced with Lenti-DLX2 OE; control, BMSCs transduced with Lenti-CTRL. d–g Analysis of the bone volume/tissue volume (BV/TV), bone mineral density (BMD), trabecular number (Tb.N), and trabecular spacing (Tb.Sp) in the respective groups. Statistical signiﬁcance was determined as described in Fig. 1; n = 6, *P < 0.05, **P < 0.01

We next checked whether Dlx2 can directly bind to the two (Fig. 5d). As shown in Fig. 5e, we found that mutation of either the predicted Dlx2-REs in the OCN promoter. To do this, we inserted upstream RE or downstream RE abrogated the ability of Dlx2 to mutated OCN promoter into the pGL3-basic plasmid in which the regulate its activity. These observations demonstrate that in MC3T3- ATTA sequence was partly mutated, generating pGL3-mut1 E1 cells, Dlx2 directly binds to the upstream RE and downstream RE (upstream RE mutant) and pGL3-mut2 (downstream RE mutant) in the OCN promoter to positively regulate its transcription.

International Journal of Oral Science (2019) 11:12 Dlx2 transactivates OCN and ALP to induce osteogenesis Zhang et al. 6 a b c Control Over 1.5 Control Over Control Over 1.5 1 000 *** *** 500 1.0 1.0

3 expression expression expression 0.5 Dlx2

2 Dlx5 0.5 Runx2 Relative fold change in fold Relative Relative fold change in fold Relative Relative fold change in fold Relative 1

0 0.0 0.0 14 21 14 21 14 21 Time/d Time/d Time/d

d Control Over e f Control Over500 Control Over 2.0 400 ** 5 300 200 1.5 100 4 100 *** 80 60 3 1.0

expression 40 expression 20 expression 2 OCN Osx 3 0.5 Relative fold change in fold Relative Msx2 1 change in fold Relative 2 Relative fold change in fold Relative 1 0 0.0 0 14 21 14 21 14 21 Time/d Time/d Time/d

ghControl Over 20 Control Over i ** 25 MC3T3-E1 15 Black Control Over 20 *** OCN 15 10 ** ** Dlx2

expression 10 gene expression

Alp 5 β 5 change in fold Relative -actin Relative fold change in fold Relative

0 0 14 21Dlx2 OCN ALP Time/d MC3T3-E1 cells Fig. 4 Dlx2 overexpression in BMSCs has no impact on Runx2, Dlx5, Msx2 and Osxexpression. RT-qPCR analysis was performed to evaluate the expression levels of Dlx2 (a), Runx2 (b), Dlx5 (c), Msx2 (d), Osx (e), OCN (F), and Alp (g) in BMSCs transduced with Lenti-DLX2 OE (over) or Lenti- CTRL (control) at 14 and 21 days after osteogenic induction. Gapdh was used as an internal control. Statistical signiﬁcance was determined as described in Fig. 1. h RT-qPCR analysis was used to evaluate the expression level of OCN and Alp upon forced overexpression of Dlx2 in MC3T3- E1 cells. i Western blot analysis was performed to measure the protein levels of OCN upon forced overexpression of Dlx2 in MC3T3-E1. β-Actin was used as an internal control. Over, BMSCs transduced with Lenti-DLX2 OE; control, BMSCs transduced with Lenti-CTRL

Dlx2 upregulates Alp expression by directly binding to its Next, we checked whether Dlx2 can bind directly to the promoter predicted Dlx2-REs (A1) in the Alp promoter. ChIP analysis was Next, we tried to investigate whether Alp was regulated by Dlx2 in performed in MC3T3-E1 cells transfected with pCMV-Dlx2-FLAG a similar way by binding to its promoter. To verify this or pCMV-FLAG, and primer set was designed to amplify the transcriptional regulation, we ﬁrst cloned and inserted the whole region A1. As shown in Fig. 6b, c, both qPCR and semi- promoter region of Alp into the pGL3-basic plasmid, generating quantitative PCR revealed that Dlx2 was highly enriched in pGL3-ALP. This vector was introduced into MC3T3-E1 cells along region A1. In addition, elevated occupancy of Pol II in the Alp with pCMV-Dlx2-FLAG to allow overexpression of Dlx2. As shown in promoter was associated with increased occupancy of Dlx2 in Fig. 6a, we observed an ~6.7-fold increase in the transcriptional this region, while the negative control (MC3T3-E1 cells intro- activity of pGL3-ALP after introduction of pCMV-Dlx2-FLAG into duced with pCMV-FLAG) showed very weak signal (data not MC3T3-E1 cells, indicating that there is at least one Dlx2-RE in the shown). Alp promoter. Supporting this notion is that bioinformatics analysis To determine whether A1 is the Dlx2-binding site in the Alp (JASPAR database) indicated that the promoter region of Alp promoter, we carried out site-directed mutagenesis in A1 contains potential Dlx2-RE (−1 194 bp to −1 187 bp, region A1). region. We inserted mutated Alp promoter into the pGL3-basic

International Journal of Oral Science (2019) 11:12 Dlx2 transactivates OCN and ALP to induce osteogenesis Zhang et al. 7 a b 30 -2 000 bp

20 OCN E1 E16

10 E2 Relative luciferase activity luciferase Relative 0 2 kb promoter pGL3-OCN + + - - IP pGL3-basic - - + + 1 2 345 678 9 10 11 12 13 14 15 16 IgG pCMV-Dlx2-FLAG + - + - Input pCMV-FLAG - + - + 1 234 5 6 789 10 11 12 13 14 15 16

c ** d 40 ** Mutation 1 Site -1 110 bp- -1 113 bp 35 -1 075 bp 30 gaaaggccttcactttaatattattagcaacatctaatttgtggacatcaagc 25 20 gaaaggccttcactttaatattattagcaacattcggcccgtggacatcaagc 20 15 Mutation 2 Site -1 444 bp- -1 447 bp

Fold enrichment Fold 10 -1 419 bp 5 ccggatattatggaggtattttctcaattgaggctttctccttccaag 0 IgG pol II 1 2 3 4 5 6 7 8 9 ccggatattatggaggtattttcctggccaaggctttctccttccaag e 30

Relative luciferase activity luciferase Relative 0 pGL3-OCN + - - + pGL3-mut1 - + - - pGL3-mut2 - - + - pCMV-Dlx2 + + + - pCMV-FLAG - - - + Fig. 5 Characterization of the mouse OCN promoter and identification of its Dlx2-response elements. a The basal luciferase activity of the whole OCN promoter construct (pGL3-OCN) and that of the empty (pGL3-basic) construct in MC3T3-E1 cells was determined by transfecting the cells with each promoter reporter construct along with the Dlx2 overexpression vector (pCMV-Dlx2-FLAG) or the empty vector (pCMV- FLAG). Cells were then harvested 24 h after the transfection, and luciferase activity was measured and normalized to the protein concentration in the cell lysate. b ChIP analysis was performed to determine the Dlx2-response elements in the OCN promoter. MC3T3-E1 cells were transfected with pCMV-Dlx2-FLAG. Semi-quantitative PCR was performed using overlapping and closely spaced primer pairs to dissect the whole OCN promoter region into 16 short (~ 175 bp) overlapping parts for identification of the bound protein. Normal IgG (2 μg) was used as control. The PCR products were then separated by electrophoresis through a 2% agarose gel. c ChIP analysis followed by RT-qPCR was performed using the same primers described in b. Statistical significance was determined as described in Fig. 1. d The sequences of the nucleotides whose sequences contain the AATT element in the OCN promoter and the sequences of two mutants. The mutant binding sites are marked in red, and the putative Dlx2-binding sites are indicated in parentheses. e The luciferase activity of wild-type OCN promoter constructs (pGL3-OCN) and the mutated ones (pGL3-mut1 and pGL3-mut2) were determined by transfecting these vectors into MC3T3-E1 cells along with pCMV-Dlx2-FLAG or pCMV-FLAG. The OCN promoter construct bearing approximately −2000 to 0 bp region was subjected to site-directed mutagenesis to substitute the AATT sequence (pGL3-OCN) with either mutation1 (pGL3-mut1) or mutation2 (pGL3-mut2). The luciferase activity was measured 24 h later plasmid, generating pGL3-mutationA1 (Fig. 6d). As shown in DISCUSSION Fig. 6e, we found that mutation of A1 significantly attenuated Dlx gene family plays a critical role in osteogenesis. Previous study the ability of Dlx2 to regulate its activity. These findings showed that Dlx2, Dlx5, and Dlx6 are upregulated in immature demonstrated that Dlx2 directly binds to region A1 in the Alp osteoblasts, while the expression of Dlx3 is elevated in differ- promoter to positively regulate its expression (Fig. 6e). Taken entiated osteoblasts and osteocytes. Dlx3 and Dlx5 potently together, all these data proved that Dlx2 promotes OCN and Alp transactivates osteoblastic marker genes15; Dlx6 also has stimu- expression by directly binding to their promoters, and then latory effects on osteogenic differentiation.17 These findings regulates osteogenic differentiation in vitro and accelerates indicate that as a member of Dlx gene family, Dlx2 may be also boneformationinvivo. involved in the osteogenic differentiation.

International Journal of Oral Science (2019) 11:12 Dlx2 transactivates OCN and ALP to induce osteogenesis Zhang et al. 8 a b ** ** 40 50

30 40 30 20 20 10

Fold enrichment Fold 10 0 Relative luciferase activity luciferase Relative + + pGL3-ALP - - 0 pGL3-basic - - + + IgG po1II A1 pCMV-Dlx2-FLAG + - + - pCMV-FLAG - + - +

cdStructure of ALP promoter Mutation A1 Site -1 187 bp to -1 194 bp A1

-1 171 bp -1 141 bp gtaaataaacagcattaattattaaaatac Input IgG RNA po1II A1 gtaaataaacagcgccggccattaaaatac

e f ** 50 Binding site 40

10 ALP mRNA

Relative luciferase activity luciferase Relative 0 + - + pGL3-ALP OCN mRNA pGL3-mutation A1 - + - pCMV-Dlx2-FLAG + + - pCMV-FLAG --+

Bone mineralization

Fig. 6 Characterization of the mouse Alp promoter and identiﬁcation of its Dlx2-response element. a The basal luciferase activity of the whole Alp promoter construct (pGL3-ALP) and that of the empty (pGL3-basic) construct in MC3T3-E1 cells was determined by transfecting the cells with each promoter reporter construct along with the Dlx2 overexpression vector (pCMV-Dlx2-FLAG) or the empty vector (pCMV-FLAG). Cells were then harvested 24 h after the transfection and luciferase activity was measured and normalized to the protein concentration in the cell lysate. b ChIP-qPCR analysis was performed to determine the Dlx2-response elements in the Alp promoter. MC3T3-E1 cells were transfected with pCMV-Dlx2-FLAG. Normal IgG (2 μg) was used as control. Pol II was used as a positive control. c ChIP analysis followed by RT-PCR was performed using the same primers described in b. d The sequence of the nucleotides whose sequences contain attaatt sequence in the Alp promoter and the sequence of the mutant. The mutant binding sites are marked in red and the putative Dlx2-binding sites are indicated in parentheses. e The luciferase activity of wild-type Alp promoter constructs (pGL3-ALP) and the mutated ones (pGL3-mutationA1) were determined by transfecting these vectors into MC3T3-E1 cells along with pCMV-Dlx2-FLAG or pCMV-FLAG. The Alp promoter construct bearing ~−2000 to 0 bp region was subjected to site-directed mutagenesis to substitute the original sequence (pGL3-ALP) with mutation (pGL3-mutationA1). The luciferase activity was measured 24 h later. f Schematic illustration of the regulation of OCN and Alp by Dlx2

Dlx2, a member of vertebrate Dlx gene family, is expressed in Here, we present evidence that forced overexpression of Dlx2 the epithelium and mesenchyme of the mandible and maxilla.18 enhances the osteogenic differentiation potential of mouse Previous studies have shown that newborn Dlx2−/− mice die BMSCs and MC3T3-E1 in vitro and accelerates bone formation immediately after birth and have abnormal craniofacial bones in vivo by directly regulating OCN and Alp. Our ﬁnding is originating from the ﬁrst branchial arch maxillary process,9 while consistent with the previous study in SCAPs that overexpression Dlx2 overexpression in cranial neural crest cell (CNCC) leads to of Dlx2 also enhanced osteogenic differentiation.19 SCAPs are premaxillary hypoplasia and spinal deformities in mice.9 Although mesenchymal-like stem cells that are able to differentiate into Dlx2 knockout and overexpression models demonstrate a crucial multiple linages, including odontoblastic and osteoblastic lineage, role of Dlx2 in promoting skeleton formation, the molecular and and do not undergo adipogenic differentiation, while BMSCs are cellular mechanisms underlying the regulation of osteogenic able to undergo osteogenic, chondrogenic, and adipogenic differentiation by Dlx2 still remain unclear. differentiation.20 Relative to BMSCs, SCAPs display elevated

International Journal of Oral Science (2019) 11:12 Dlx2 transactivates OCN and ALP to induce osteogenesis Zhang et al. 9

Table 1. Primers used in this study

Primer Sequence Details Reference

Primers for RT-qPCR ep Dlx2 FW CATGGGCTCCTACCAGTACCAC qPCR primer for Dlx2 8 ep Dlx2 RV TCGGATTTCAGGCTCAAGGTC qPCR primer for Dlx2 8 ep Gapdh FW GGTGAAGGTCGGTGTGAACG qPCR primer for Gapdh 8 ep Gapdh RV CTCGCTCCTGGAAGATGGTG qPCR primer for Gapdh 8 ep ALP FW TGGGCATTGTGACTACCACTCGG qPCR primer for Alp 8 ep ALP RV CCTCTGGTGGCATCTCGTTATCC qPCR primer for Alp 8 ep OCN FW GGACCATCTTTCTGCTCACTCTG qPCR primer for Ocn 8 ep OCN RV GTTCACTACCTTATTGCCCTCCTG qPCR primer for Ocn 8 ep Runx2 FW AACTTCCTGTGCTCCGTGCTG qPCR primer for Runx2 8 ep Runx2 Rv TCGTTGAACCTGGCTACTTGG qPCR primer for Runx2 8 ep Msx2 FW GGAGCACCGTGGATACAGGA qPCR primer for Msx2 8 ep Msx2 RV AGGCTAGAAGCTGGGATGTGG qPCR primer for Msx2 8 ep OSX FW CCCTTCTCAAGCACCAATGG qPCR primer for Osx 41 ep OSX RV AAGGGTGGGTAGTCATTTGCATA qPCR primer for Osx 41 Primers for ChIP analysis Primer-F16 gaggctgagagagagagcaca For ChIP-qPCR This study Primer-R16 tgctgtggtaggtgattgc For ChIP-qPCR This study Primer-F15 gcagacactgaaaatcacagg For ChIP-qPCR This study Primer-R15 tgtgctctctctctcagcctc For ChIP-qPCR This study Primer-F14 cataaaactaaccaggacactcc For ChIP-qPCR This study Primer-R14 gcctgtgattttcagtgtctgc For ChIP-qPCR This study Primer-F13 ccacaatgggctaggctct For ChIP-qPCR This study Primer-R13 catccagtgggggtgtgt For ChIP-qPCR This study Primer-F12 acacaagcagggctagaacc For ChIP-qPCR This study Primer-R12 agcccattgtgggagctac For ChIP-qPCR This study Primer-F11 ttgtggacatcaagcggg For ChIP-qPCR This study Primer-R11 aggttctagccctgcttgtg For ChIP-qPCR This study Primer-F10 tttcatttccacctagagcaag For ChIP-qPCR This study Primer-R10 tcgggggtatctggtttcag For ChIP-qPCR This study Primer-F9 gtttggcatggagcctttg For ChIP-qPCR This study Primer-R9 tgtgttacagtcacttgctctagg For ChIP-qPCR This study Primer-F8 gccctcttctagtgtgtctgaa For ChIP-qPCR This study Primer-R8 cccaagttcaaaggctccat For ChIP-qPCR This study Primer-F7 cacagttggactggggaggt For ChIP-qPCR This study Primer-R7 cttcagacacactagaagagggc For ChIP-qPCR This study Primer-F6 tcaccagcgcaaatcacac For ChIP-qPCR This study Primer-R6 ctgtgtgctttcttattcacctacc For ChIP-qPCR This study Primer-F5 aagggaaacaccacccactc For ChIP-qPCR This study Primer-R5 atttgcgctggtgaggac For ChIP-qPCR This study Primer-F4 cctccactgcctcaagaact For ChIP-qPCR This study Primer-R4 aaggctggaaaggagtgggt For ChIP-qPCR This study Primer-F3 acagaaggcaggtcagctaca For ChIP-qPCR This study Primer-R3 gggaagagtaaggtggaggtt For ChIP-qPCR This study Primer-F2 ttccccctagccgaacaag For ChIP-qPCR This study Primer-R2 acctgccttctgtgatcctct For ChIP-qPCR This study Primer-F1 agaggatcacagaaggcaggt For ChIP-qPCR This study Primer-R1 aaaggatgctgtggttggtgattg For ChIP-qPCR This study Primer-A1F tgcctgggtttgttttcatttgt For ChIP-qPCR This study Primer-A1R caagggaaatttcctagcac For ChIP-qPCR This study

International Journal of Oral Science (2019) 11:12 Dlx2 transactivates OCN and ALP to induce osteogenesis Zhang et al. 10 secretion of proteins involved in metabolic processes, chemo- endochondral ossification.34 Our previous study has shown that kines, and neutrophins, whereas BMSCs secret much more Dlx2 is involved in endochondral ossification.9,35 Therefore, Dlx2 is proangiogenic factors and ECM proteins. Therefore, overexpres- also involved in the bone formation of lower jaw, but is not as sion of Dlx2 in SCAPs and BMSCs could be used in dentin important as Dlx5. Together, both Dlx2 and Dlx5 are involved in regeneration and bone formation, respectively.21 the development of mandible and maxilla. Considering the fact The rodent Dlx5 and Dlx2 HD transcription factors are critical that a variety of homeobox genes work coordinately during the for bone development. During osteogenesis, Dlx5 upregulates Alp bone formation,36 further investigations are required to find out expression while suppresses OCN transcription.22 On the contrary, how Dlx2 is involved during the bone formation in mandible and we found that Dlx2 transactivates OCN and ALP by directly maxilla. binding to their promoters. Moreover, we identified two Dlx2-REs In conclusion, our data demonstrated for the first time that in OCN promoter, and one Dlx2-RE in Alp promoter involved in forced overexpression of Dlx2 enhances the osteogenic differ- Dlx2-induced Alp expression. These findings demonstrate that entiation potential of BMSCs and MC3T3-E1 cells by directly Dlx2 is a crucial regulator regulating the osteogenic differentia- upregulating OCN and Alp (Fig. 6f). We also presented evidence tion potential of both mesenchymal stem cells and osteogenic that there are Dlx2-REs in mouse OCN and Alp promoter that cells. mediate the regulation of Dlx2 on OCN or Alp expression. This BMP-2 is one of the most important cytokines promoting study may present a promising future strategy for the treatment of differentiation of mesenchymal cells into osteoblasts.23 Stimulated bone defects with Dlx2-overexpressing BMSCs. by BMP-2, the transcription factors Dlx5, Msx2, and Runx2 work coordinately to regulate osteogenic differentiation.24 Both Runx2 and Dlx3 positively regulate OCN expression while Dlx5 represses MATERIAL AND METHODS its expression; Dlx5 stimulates ALP expression, whereas Msx2 Isolation and culture of mouse BMSCs depresses its expression.25 At the onset of osteogenic differentia- All animal experiments were performed according to guidelines of tion, Msx2 is released from the promoter of OCN, while Dlx3, Dlx5, the Institutional Animal Care and Use Committees of the Shanghai and Runx2 are recruited. The released Msx2 then binds to the Alp Ninth People’s Hospital, Shanghai Jiao Tong University School of promoter to upregulate its expression. At a later stage of Medicine. All experimental protocols were reviewed and approved osteogenic differentiation, during matrix mineralization, Dlx5 by the Institutional Animal Care and Use Committees of the replaces Dlx3 to regulate OCN expression.26 Runx2 and Osx are Shanghai Ninth People’s Hospital, Shanghai Jiao Tong University another key transcription factors that are necessary for osteogen- School of Medicine, Shanghai, China. BMSCs were isolated from esis.27 After differentiating into pre-osteoblasts, Runx2 and Osx the tibias of 6-week-old male C57/BL6 mice ((10 ± 0.5) g) and promote the cells to produce bone matrix. Dlx gene family is cultured according to previous study.37 involved in the regulation of Runx2 and OSX transcription. Dlx5 An MC3T3-E1 cell line was obtained from the Cell Bank of the induces expression of Runx2 and Osx, which work sequentially to Chinese Academy of Science (Shanghai, China), C3H10 T1/2 cell stimulate the expression of OCN and Alp. Dlx3 also contributes to line from American Type Culture Collection (Rockville, MD, USA), the activation of Runx2 during osteogenic differentiation.28 These and a Human embryonic kidney 293T (HEK 293T) cell line from findings indicate that Dlx gene family plays a crucial role in American Type Culture Collection (Rockville, MD, USA). The expression of osteogenic-associated genes. The results of the MC3T3-E1, C3H10 T1/2 cell line, and HEK 293T cells were cultured present study showed that overexpression of Dlx2 showed no as described previously.38,39 OIM contained 50 mg· L−1 ascorbic significant effects on Runx2, Msx2, and Dlx5 expression upon acid, 1 × 10−7 mol· L−1 dexamethasone, and 50 mg· L−1 β- osteogenic induction, but stimulated OCN and Alp expression, glycerophosphate plus α-MEM (Sigma-Aldrich Corp. (St. Louis, indicating that Runx2, Msx2, and Dlx5 may not participate in Dlx2- MO, USA)). HEK 293T cell line was utilized for packaging viral induced osteogenesis; Dlx2 may directly upregulate OCN and Alp constructs. to promote osteogenic differentiation. Alp and OCN are two key marker genes of osteoblastic cells. ALP Lentiviral construction and transduction plays a critical role in early osteogenesis and hydrolyzes various The lentiviral expression system overexpressing Dlx2 was termed types of phosphates to promote cell maturation and calcification, as Lenti-Dlx2 OE. The open reading frame of mice Dlx2 while OCN promotes the later stage of osteogenesis through (NM_010054) was synthesized and cloned into pL/IRES/GFP combining with minerals.29,30 Both Dlx3 and Dlx5 directly plasmid (Novobio, Shanghai, China) for generating pL/IRES/GFP- upregulate Alp expression, while OCN is activated by Dlx3 but DLX2. The empty lentiviral expression system without insertion suppressed by Dlx5.16,31 Moreover, forced overexpression of Dlx5 was termed as Lenti-CTRL and used as the control. 293T cells were in BMSCs led to a reduction in the mineralized matrix deposition, then transfected with plasmids listed above. The transfection and and impaired the ability of these cells to develop to the final lentiviral transduction was done as described previously.40,41 stages of osteogenesis, and severely affected in vivo bone formation in immunodeficient mice.13 Although Dlx2 shares a ALP, Alizarin red staining, and semi-quantitative analysis strong sequence similarity with Dlx5, we found that Dlx2 positively Transduced BMSCs or MC3T3-E1 cells were first cultured in OIM for regulates both Alp and OCN expression in BMSCs and MC3T3-E1 14 or 21 days. ALP and Alizarin red staining were carried out as cells. described previously.40 ALP staining was carried out with BCIP/ A previous study showed that loss of Dlx1/2−/− leads to NBT Alkaline Phosphatase Color Development Kit (Beyotime abnormal bone formation of the upper jaw, while Dlx5/6−/− Institute of Biotechnology, China), and semi-quantitative analysis deficient mice exhibit profound abnormalities of the lower jaw of ALP activity was performed using p-nitrophenyl phosphate (p- tissue.22 However, piles of studies revealed that Dlx5 is the master NPP) (Sigma-Aldrich) as the substrate. For Alizarin red staining, regulator of osteogenic differentiation, since it directly controls cells were first fixed with 70% ethanol. Afterward, the fixed cells the transcription of multiple osteogenic-associated genes, includ- were stained with 2% Alizarin Red (Sigma-Aldrich), according to ing Alp, OCN, Runx2, OSX, and Smads family, affecting the whole the previous study.19 process of bone formation.32,33 Therefore, Dlx5 is also involved in the development of the upper jaw, but may be not as important Semi-quantitative RT-PCR, RT-qPCR, and Western blot analysis as Dlx2. On the other hand, the maxilla is only formed by RT-PCR, RT-qPCR, and western blot analysis were performed as intramembranous ossification of the craniofacial mesenchyme, described previously.40 Total RNA was extracted from cultured while the mandible can be formed by both intramembranous and cells using TRIzol RNA Isolation reagent (Takara, Tokyo, Japan),

International Journal of Oral Science (2019) 11:12 Dlx2 transactivates OCN and ALP to induce osteogenesis Zhang et al. 11 according to the manufacturer’s instruction. Three independent purified with phenol–chloroform extraction. ChIP samples were cultures were used for RNA preparations. First-strand cDNA was then analyzed with RT-PCR and RT-qPCR. Pol II was used as a generated with High-Capacity cDNA Reverse Transcription Kit, positive control, as described in a previous study.15 The primers (Applied Biosystems, San Diego, CA), and one microliter of each RT used in ChIP analysis are listed in Table 1. reaction mixture was amplified with Ex Taq DNA polymerase (Takara, Tokyo, Japan). As for RT-qPCR, cDNA was amplified using premix SYBR Green Ex Taq reagent kit (DRR820A, Takara) with a ACKNOWLEDGEMENTS STEP ONE PLUS real-time PCR system (Applied Biosystems, Forster The authors thank Dr. Wenbin Zhang, Dr. Jianjun Zhang, Dr. Ming Yan and Dr. Xu City, CA), according to the manufacturer’s instruction. All the Wang from the Shanghai Ninth People’s Hospital for helping us with the technique. primers used in this study are listed in Table 1. As for western We would like to thank the members of the Shen- and Chen-labs for helpful advice blotting, anti-Dlx2 (1:800; ab85995, Abcam, Cambridge, UK), anti- and discussion. We also thank Native EE for its linguistic assistance during the OCN (1:1 000; ab93876, Abcam), and anti-β-actin (1:3 000; preparation of this manuscript. This work is supported by grant (81771036) from National Natural Science Foundation of China (to S.G.S.), grant (81741028) from EPR16769, Abcam, Cambridge, UK) were used for the detection National Natural Science Foundation of China (to J.D.) and grant (17410710500) of Dlx2, OCN, and β-actin, respectively. The secondary antibodies Shanghai International Scientific and Technological Cooperation Projects Laser Micro- used this study were bought from Sigma-Aldrich and conjugated machine and Vascularization of TCP/PCL Scaffolds (to W.Z.). to horseradish peroxidase (anti-rabbit (1:5 000, A0545) or anti- mouse (1:5 000, SAB3701214)). ADDITIONAL INFORMATION In vivo osteogenic differentiation The online version of this article (https://doi.org/10.1038/s41368-019-0046-1) The osteogenic differentiation potential of transduced BMSCs was contains supplementary material, which is available to authorized users. evaluated with in vivo ectopic bone formation analysis, as described previously.13 Briefly, osteogenic-induced cells of Competing interests: The authors declare no competing interests. passage 3 were injected into the β-TCP (Shanghai Rebone Biomaterials Co., Ltd., Shanghai, China) with a syringe, and REFERENCES appropriate volume of OIM was added to cover the BMSCs/ β-TCP 1. Ulsamer, A. et al. BMP-2 induces Osterix expression through up-regulation of Dlx5 constructs. After 7 days of culturing in vitro, the constructs were – implanted subcutaneously in the back of immunocompromised and its phosphorylation by p38. J. Biol. Chem. 283, 3816 3826 (2008). 2. Stock, D. W. et al. The evolution of the vertebrate Dlx gene family. Proc. Natl Acad. female nude mice (CD-1 Nu/Nu, 10-week-old, Charles River). At Sci. USA 93, 10858–10863 (1996). each time point (6 and 8 weeks after implantation), six mice for 3. Lee, M.-H., Kwon, T.-G., Park, H.-S., Wozney, J. M. & Ryoo, H.-M. BMP-2-induced each group were sacrificed with an overdose of pentobarbital Osterix expression is mediated by Dlx5 but is independent of Runx2. Biochem. (210 mg/kg intraperitoneally) to retrieve the BMSCs/β-TCP con- Biophys. Res. Commun. 309, 689–694 (2003). structs. Micro-CT images of each cellular construct were taken 4. Shirakabe, K., Terasawa, K., Miyama, K., Shibuya, H. & Nishida, E. Regulation of the with a micro-CT system (SMX-100CT-SV3; Shimadzu, Japan). activity of the transcription factor Runx2 by two homeobox proteins, Msx2 and Radiological density of each cellular construct was measured in Dlx5. Genes Cells 6, 851–856 (2001). fi Hounsfield density units. 5. Depew, M. J., Lufkin, T. & Rubenstein, J. L. Speci cation of jaw subdivisions by Dlx genes. Science 298, 381–385 (2002). 6. Ferguson, C. A., Tucker, A. S. & Sharpe, P. T. Temporospatial cell interactions Luciferase assay regulating mandibular and maxillary arch patterning. Development 127, 403–412 The whole mouse OCN promoter (2000 bp) was subcloned into (2000). the pGL3 basic vector (Promega), generating pGL3-OCN. To 7. Merlo, G. R. et al. Multiple functions of Dlx genes. Int. J. Dev. Biol. 44, 619–626 produce constructs that contain mutation in E6 region, we carried (2004). out mutagenic PCR with Mt1 primer and Mt2 primer, generating 8. Sun, H., Liu, Z., Li, B., Dai, J. & Wang, X. Effects of DLX2 overexpression on pGL3-mut1. Likewise, Mt3 and Mt4 primers were used for the osteogenic differentiation of MC3T3‑E1 cells. Exp. Ther. Med. 9, 2173–2179 generating pGL3-mut2 (Table 1). The open reading frame of mice (2015). Dlx2 (NM_010054) was synthesized and cloned into pCMV-FLAG, 9. Dai, J. et al. The effect of overexpression of Dlx2 on the migration, proliferation and osteogenic differentiation of cranial neural crest stem cells. Biomaterials 34, generating pCMV-Dlx2-FLAG for Dlx2 overexpressing. Transfection – 31 1898 1910 (2013). assay was performed as described previously. As for each 10. Pittenger, M. F. et al. Multilineage potential of adult human mesenchymal stem transfection assay, 0.5 μg of the Dlx2 overexpression vector cells. Science 284, 143–147 (1999). (pCMV-Dlx2-FLAG) or pCMV-FLAG and 0.5 μg of the luciferase 11. Kuznetsov, S. A. et al. Single‐colony derived strains of human marrow stromal reporter vector were transfected into MC3T3-E1 cells. Similarly, we fibroblasts form bone after transplantation in vivo. J. Bone Mineral. Res. 12, inserted the whole Alp promoter into the pGL3-basic vectors, 1335–1347 (1997). generating pGL3-ALP, and produced a vector that contains 12. Krebsbach, P. H. et al. Bone formation in vivo: comparison of osteogenesis by mutated Alp promoter, generating pGL3-mutationA1. Afterward, transplanted mouse and human marrow stromal fibroblasts. Transplantation 63, – pCMV-Dlx2-FLAG or pCMV-FLAG was introduced into MC3T3-E1 1059 1069 (1997). cells along with pGL3-ALP or pGL3-mutationA1. 13. Muraglia, A. et al. DLX5 overexpression impairs osteogenic differentiation of human bone marrow stromal cells. Eur. J. Cell Biol. 87, 751–761 (2008). 14. Dalby, M. J. et al. The control of human mesenchymal cell differentiation using ChIP analysis nanoscale symmetry and disorder. Nat. Mater. 6, 997 (2007). 15 ChIP analysis was carried out according to the standard protocol. 15. Hassan, M. Q. et al. Dlx3 transcriptional regulation of osteoblast differentiation: MC3T3-E1 cells transfected with pCMV-Dlx2-FLAG or pCMV-FLAG temporal recruitment of Msx2, Dlx3, and Dlx5 homeodomain proteins to chro- were fixed with 1% formaldehyde and this cross-linking was matin of the osteocalcin gene. Mol. Cell. Biol. 24, 9248–9261 (2004). quenched with glycine (0.125 mol· L−1 for final concentration). 16. Ryoo, H.-M. et al. Stage-specific expression of Dlx-5 during osteoblast differ- After the cells were lysed and homogenized with a Dounce entiation: involvement in regulation of osteocalcin gene expression. Mol. Endo- – homogenizer, the nuclei were collected by centrifuging. The crinol. 11, 1681 1694 (1997). nuclei pellet was then resuspended in sonication buffer followed 17. Pfeffer, U. et al. The coding region of the human DLX6 gene contains a poly- morphic CAG/CCG repeat. Int. J. Oncol. 18, 1293–1297 (2001). by sonication. Chromatin were incubated with anti-FLAG anti- 18. Depew, M. J., Simpson, C. A., Morasso, M. & Rubenstein, J. L. R. Reassessing the Dlx bodies (F3165, Sigma-Aldrich), mouse IgG and anti-RNA- code: the genetic regulation of branchial arch skeletal pattern and development. polymerase II (PLA0292, Sigma-Aldrich), and this immune complex J. Anat. 207, 501–561 (2005). were incubated with protein G sepharose (17-0618-01, GE 19. Qu, B. et al. Distal-less homeobox 2 promotes the osteogenic differentiation Amersham). The DNA fragments in this immune complex were potential of stem cells from apical papilla. Cell Tissue Res. 357, 133–143 (2014).

International Journal of Oral Science (2019) 11:12 Dlx2 transactivates OCN and ALP to induce osteogenesis Zhang et al. 12 20. Huang, G. T. J. et al. Stem/progenitor cell–mediated de novo regeneration of 34. Tortelli, F., Tasso, R., Loiacono, F. & Cancedda, R. The development of tissue- dental pulp with newly deposited continuous layer of dentin in an in vivo model. engineered bone of different origin through endochondral and intramem- Tissue Eng. Part A 16, 605–615 (2009). branous ossification following the implantation of mesenchymal stem cells and 21. Yu, S., Zhao, Y., Ma, Y. & Ge, L. Profiling the secretome of human stem cells from osteoblasts in a murine model. Biomaterials 31, 242–249 (2010). dental apical papilla. Stem Cells Dev. 25, 499–508 (2016). 35. Zhang, J., Zhang, W., Shi, J., Dai, J. & Shen, S. G. Dlx2 overexpression enhanced 22. Samee, N., De Vernejoul, M.-C. & Levi, G. Role of DLX regulatory proteins in accumulation of type II collagen and aggrecan by inhibiting MMP13 expression osteogenesis and chondrogenesis. Crit. Rev. Eukaryot. Gene Expr. 17, 173–186 in mice chondrocytes. Biochem. Biophys. Res. Commun. 503, 528–535 (2018). (2007). 36. Thesleff, I. Homeobox genes and growth factors in regulation of craniofacial and 23. Hanai, J.-i et al. Interaction and functional cooperation of PEBP2/CBF with smads tooth morphogenesis. Acta Odontol. Scand. 53, 129–134 (1995). synergistic induction of the immunoglobulin germline Cα promoter. J. Biol. Chem. 37. Russo, F. P. et al. The bone marrow functionally contributes to liver fibrosis. 274, 31577–31582 (1999). Gastroenterology 130, 1807–1821 (2006). 24. Nishimura, R., Hata, K., Harris, S., Ikeda, F. & Yoneda, T. Core-binding factor α1 38. Yamamoto, K., Arakawa, T., Ueda, N. & Yamamoto, S. Transcriptional roles of (Cbfa1) induces osteoblastic differentiation of C2C12 cells without interactions nuclear factor B and nuclear factor-interleukin-6 in the tumor necrosis factor- with Smad1 and Smad5. Bone 31, 303–312 (2002). dependent induction of cyclooxygenase-2 in MC3T3-E1 cells. J. Biol. Chem. 270, 25. Javed, A. et al. runt homology domain transcription factors (Runx, Cbfa, and AML) 31315–31320 (1995). mediate repression of the bone sialoprotein promoter: evidence for promoter 39. Heim, M. et al. Phytanic acid, a natural peroxisome proliferator-activated receptor context-dependent activity of Cbfa proteins. Mol. Cell. Biol. 21, 2891–2905 (2001). (PPAR) agonist, regulates glucose metabolism in rat primary hepatocytes. FASEB J. 26. Langenbach, F. & Handschel, J. Effects of dexamethasone, ascorbic acid and β- 16, 718–720 (2002). glycerophosphate on the osteogenic differentiation of stem cells in vitro. Stem 40. Xie, Q. et al. The role of miR-135-modified adipose-derived mesenchymal stem Cell Res. Ther. 4, 117 (2013). cells in bone regeneration. Biomaterials 75, 279–294 (2016). 27. Zhang, C. Transcriptional regulation of bone formation by the osteoblast-specific 41. Cheng, S. L. et al. MSX2 promotes osteogenesis and suppresses adipogenic dif- transcription factor Osx. J. Orthop. Surg. Res. 5, 37 (2010). ferentiation of multipotent mesenchymal progenitors[J]. J. Biol. Chem. 278, 28. Hassan, M. Q. et al. BMP2 commitment to the osteogenic lineage involves acti- 45969–45977 (2003). vation of Runx2 by DLX3 and a homeodomain transcriptional network. J. Biol. Chem. 281, 40515–40526 (2006). 29. Glowacki, J., Rey, C., Glimcher, M., Cox, K. & Lian, J. A role for osteocalcin in Open Access This article is licensed under a Creative Commons osteoclast differentiation. J. Cell. Biochem. 45, 292–302 (1991). Attribution 4.0 International License, which permits use, sharing, 30. Beck, G. R., Zerler, B. & Moran, E. Phosphate is a specific signal for induction adaptation, distribution and reproduction in any medium or format, as long as you give of osteopontin gene expression. Proc. Natl Acad. Sci. USA 97, 8352–8357 appropriate credit to the original author(s) and the source, provide a link to the Creative (2000). Commons license, and indicate if changes were made. The images or other third party 31. Kim, Y.-J., Lee, M.-H., Wozney, J. M., Cho, J.-Y. & Ryoo, H.-M. Bone morphogenetic material in this article are included in the article’s Creative Commons license, unless protein-2-induced alkaline phosphatase expression is stimulated by Dlx5 and indicated otherwise in a credit line to the material. If material is not included in the repressed by Msx2. J. Biol. Chem. 279, 50773–50780 (2004). article’s Creative Commons license and your intended use is not permitted by statutory 32. Lee, M.-H. et al. BMP-2-induced Runx2 expression is mediated by Dlx5, and TGF- regulation or exceeds the permitted use, you will need to obtain permission directly β1 opposes the BMP-2-induced osteoblast differentiation by suppression of Dlx5 from the copyright holder. To view a copy of this license, visit http://creativecommons. expression. J. Biol. Chem. 278, 34387–34394 (2003). org/licenses/by/4.0/. 33. Robledo, R. F., Rajan, L., Li, X. & Lufkin, T. The Dlx5 and Dlx6 homeobox genes are essential for craniofacial, axial, and appendicular skeletal development. Genes Dev. 16, 1089–1101 (2002). © The Author(s) 2019

International Journal of Oral Science (2019) 11:12